Objective: To explore the influencing factors of fertilization failure (FF) during in vitro fertilization (IVF) to prevent and manage it in clinical practice and improve treatment efficiency.
Methods: IVF cycles were included and grouped according to the fertilization rate. There were 75 cycles with a fertilization rate of 0, i.e., complete FF, and 98 cycles with a fertilization rate of <30%, i.e., partial FF, and these cycles were included in the FF group; and there were 2301 cycles with a fertilization rate of ≥30%, and included in the normal fertilization(NF) group. Sperm quality of males, basic conditions of females, clinical ovulation induction and laboratory fertilization were compared between the two groups, and no differences were observed. Multivariate logistic regression analysis was performed using FF in the IVF process as the dependent variable, and the indicators with statistically significant differences in the univariate analysis as independent variables to screen the independent risk factors for FF in IVF.
Results: There were significant differences in female age, infertile duration, initial dose of Gn, Gn dose/egg, sperm concentration before treatment, sperm motility, percentage of normal sperm morphology, sperm concentration after treatment, and fertilization concentration(10,000 sperms/ml), and the differences were statistically significant(all P<0.05). Multivariate logistic regression analysis showed that a high percentage of primary infertility, a low percentage of tubal factors, a low percentage of normal sperm morphology, and low sperm concentration after treatment were independent risk factors for FF, and the differences were statistically significant (all P<0.05). Logistic binary regression fitting was used to construct a ROC curve prediction model for combined prediction of fertilization failure using various indicators, and the AUC was 74.6%.
Conclusion: A high percentage of primary infertility, a low percentage of tubal factors, a low percentage of normal sperm morphology, and low sperm concentration after treatment are independent risk factors for FF. The ROC curve model using combined indicators to predict FF constructed by logistic binary regression fitting is valuable in FF prediction.